As defined by the World Health Organization, an adverse drug reaction (ADR) is an effect that is "noxious and unintended, and which occurs at doses used in man for prophylaxis, diagnosis, or therapy." The term has since been changed to an adverse drug event (ADE) which is more comprehensive and reflects any injury resulting from a pharmaceutical intervention.

Adverse drug events comprise a wide range of effects but in general can be designated as significant or non-significant. Significant effects again denote a range of effects, from temporary or permanent injury or death, to effects that, while not life-threatening, still require substantive intervention. The interventions may include further treatments (i.e. adding a drug) or may involve testing and procedures. Non-significant ADEs are minor effects that require, for instance, discontinuation of a medication. A non-significant effect would be mild dyspepsia that resolves spontaneously after stopping a non-steroidal anti-inflammatory drug (NSAID) (Ibuprofen or Motrin).

The standard method used for reducing the incidence of potential and actual ADEs in the VA is on-site pharmacist and physician review. However, due to personnel, time and information constraints, a substantial number of preventable ADEs still occur. To assist in identifying potential ADEs, various computer software programs have been developed to help profile patients' drug-related risks but as yet none have been shown to influence clinical outcomes in the ambulatory setting. We pilot tested a computer alert program and used it to "profile" a small group of patients at high risk for a drug-related adverse outcome. Relevant alerts were sent to patients' primary and principal physicians. In a pre-post evaluation, we found that several surrogate markers of ADEs improved and that hospitalizations decreased. The purpose of the proposed study is to evaluate whether the methodology of patient-profiling and provider feedback truly improves quality. Specifically, we hypothesized that patient risk profiling (detailing potential ADEs) combined with provider feedback would improve clinical outcomes as measured by reducing actual ADEs and by decreasing health care utilization. To test this hypothesis, we are conducting a randomized controlled trial. At this time, we have completed the random assignment of approximately 925 patients to either at-risk profiling with provider feedback (intervention) or to the current standard of care of pharmacist and physician review (control). We are presently abstracting data from the medical records of those patients who are 1-year post-randomization. The data will be used to compare the number of actual ADEs, as measured by trained pharmacist reviewers blinded to patient assignment. We are also in the planning stages of using health utilization and economic outcomes, as measured by Decision Support Services (DSS). Finally, we have conducted a baseline survey (paper accepted by Medical Care, in press) and are now implementing a post-intervention survey to better understand providers’ attitudes towards and knowledge of drug interactions and drug alerts.

Genders Eligible for Study:  Both

Criteria

• Patient must be alive
• Patient must be currently active in GLA system
• Medication for which alert was generated must be currently active
• Patient provider must not be Peter Glassman